yetangitu/recoll
1
0
Fork 0
Code Issues Pull requests Releases Wiki Activity

Arrange for plaintorich to keep track of the link between match area and search terms. Use this in preview to allow walking the matches for a given search term group

Browse source
This commit is contained in:
Jean-Francois Dockes 2012-08-18 12:15:43 +02:00
parent 35ed2b3ca9
commit 1ed0f4ddf8
8 changed files with 291 additions and 156 deletions
Download patch file Download diff file Expand all files Collapse all files

28
src/kde/kioslave/kio_recoll/htmlif.cpp
View file

@ -204,24 +204,32 @@ void RecollProtocol::queryDetails()
class PlainToRichKio : public PlainToRich {
public:
PlainToRichKio(const string& nm)
: PlainToRich() , m_name(nm)
: m_name(nm)
{
}
virtual ~PlainToRichKio() {}
virtual string header() {
if (m_inputhtml) {
return snull;
return cstr_null;
} else {
return
string("<html><head>"
return string("<html><head>"
"<META http-equiv=\"Content-Type\""
"content=\"text/html;charset=UTF-8\"><title>") +
m_name +
string("</title></head><body><pre>");
"content=\"text/html;charset=UTF-8\"><title>").
append(m_name).
append("</title></head><body><pre>");
}
}
virtual string startMatch() {return string("<font color=\"blue\">");}
virtual string endMatch() {return string("</font>");}
virtual string startMatch(unsigned int)
{
return string("<font color=\"blue\">");
}
virtual string endMatch()
{
return string("</font>");
}
const string &m_name;
};

238
src/qtgui/preview_w.cpp
View file

@ -68,54 +68,130 @@ using std::pair;
// Subclass plainToRich to add <termtag>s and anchors to the preview text
class PlainToRichQtPreview : public PlainToRich {
public:
int lastanchor;
PlainToRichQtPreview()
: m_curanchor(1), m_lastanchor(0)
{
lastanchor = 0;
}
virtual ~PlainToRichQtPreview() {}
virtual string header() {
bool haveAnchors()
{
return m_lastanchor != 0;
}
virtual string header()
{
if (m_inputhtml) {
return snull;
return cstr_null;
} else {
if (prefs.previewPlainPre) {
m_eolbr = false;
return string("<qt><head><title></title></head><body>"
"<pre>");
// Note we could also use the following for line-folding instead of <br>s
// This would be possible without recomputing the whole text, much better perfs
// for toggling wrap/no-wrap
// "<pre style=\"white-space: pre-wrap\">");
// Note: we could also use the following for
// line-folding instead of <br>s This would be
// possible without recomputing the whole text, much
// better perfs for toggling wrap/no-wrap:
// <pre style=\"white-space: pre-wrap\">
} else {
m_eolbr = true;
return string("<qt><head><title></title></head><body>");
}
}
}
virtual string startMatch()
virtual string startMatch(unsigned int grpidx)
{
return string("<span style='color: ")
+ string((const char *)(prefs.qtermcolor.toUtf8()))
+ string(";font-weight: bold;")
+ string("'>");
LOGDEB2(("startMatch, grpidx %u\n", grpidx));
grpidx = m_hdata->grpsugidx[grpidx];
LOGDEB2(("startMatch, ugrpidx %u\n", grpidx));
m_groupanchors[grpidx].push_back(++m_lastanchor);
m_groupcuranchors[grpidx] = 0;
return string("<span style='color: ").
append((const char *)(prefs.qtermcolor.toUtf8())).
append(";font-weight: bold;").
append("'>").
append("<a name=\"").
append(termAnchorName(m_lastanchor)).
append("\">");
}
virtual string endMatch() {return string("</span>");}
virtual string termAnchorName(int i) {
virtual string endMatch()
{
return string("</a></span>");
}
virtual string termAnchorName(int i) const
{
static const char *termAnchorNameBase = "TRM";
char acname[sizeof(termAnchorNameBase) + 20];
sprintf(acname, "%s%d", termAnchorNameBase, i);
if (i > lastanchor)
lastanchor = i;
return string(acname);
}
virtual string startAnchor(int i) {
return string("<a name=\"") + termAnchorName(i) + "\">";
virtual string startChunk()
{
return "<pre>";
}
virtual string endAnchor() {
return string("</a>");
int nextAnchorNum(int grpidx)
{
LOGDEB2(("nextAnchorNum: group %d\n", grpidx));
map<unsigned int, unsigned int>::iterator curit =
m_groupcuranchors.find(grpidx);
map<unsigned int, vector<int> >::iterator vecit =
m_groupanchors.find(grpidx);
if (grpidx == -1 || curit == m_groupcuranchors.end() ||
vecit == m_groupanchors.end()) {
if (m_curanchor >= m_lastanchor)
m_curanchor = 1;
else
m_curanchor++;
} else {
if (curit->second >= vecit->second.size() -1)
m_groupcuranchors[grpidx] = 0;
else
m_groupcuranchors[grpidx]++;
m_curanchor = vecit->second[m_groupcuranchors[grpidx]];
LOGDEB2(("nextAnchorNum: curanchor now %d\n", m_curanchor));
}
virtual string startChunk() { return "<pre>";}
return m_curanchor;
}
int prevAnchorNum(int grpidx)
{
map<unsigned int, unsigned int>::iterator curit =
m_groupcuranchors.find(grpidx);
map<unsigned int, vector<int> >::iterator vecit =
m_groupanchors.find(grpidx);
if (grpidx == -1 || curit == m_groupcuranchors.end() ||
vecit == m_groupanchors.end()) {
if (m_curanchor <= 1)
m_curanchor = m_lastanchor;
else
m_curanchor--;
} else {
if (curit->second <= 0)
m_groupcuranchors[grpidx] = vecit->second.size() -1;
else
m_groupcuranchors[grpidx]--;
m_curanchor = vecit->second[m_groupcuranchors[grpidx]];
}
return m_curanchor;
}
QString curAnchorName() const
{
return QString::fromUtf8(termAnchorName(m_curanchor).c_str());
}
private:
int m_curanchor;
int m_lastanchor;
// Lists of anchor numbers (match locations) for the term (groups)
// in the query (the map key is and index into HighlightData.groups).
map<unsigned int, vector<int> > m_groupanchors;
map<unsigned int, unsigned int> m_groupcuranchors;
};
void Preview::init()
@ -141,8 +217,24 @@ void Preview::init()
QHBoxLayout *layout3 = new QHBoxLayout(0);
searchLabel = new QLabel(this);
layout3->addWidget(searchLabel);
searchTextLine = new QLineEdit(this);
layout3->addWidget(searchTextLine);
searchTextCMB = new QComboBox(this);
searchTextCMB->setEditable(true);
searchTextCMB->setInsertPolicy(QComboBox::NoInsert);
searchTextCMB->setDuplicatesEnabled(false);
for (unsigned int i = 0; i < m_hData.ugroups.size(); i++) {
QString s;
for (unsigned int j = 0; j < m_hData.ugroups[i].size(); j++) {
s.append(QString::fromUtf8(m_hData.ugroups[i][j].c_str()));
if (j != m_hData.ugroups[i].size()-1)
s.append(" ");
}
searchTextCMB->addItem(s);
}
searchTextCMB->setEditText("");
layout3->addWidget(searchTextCMB);
nextButton = new QPushButton(this);
nextButton->setEnabled(TRUE);
layout3->addWidget(nextButton);
@ -160,7 +252,7 @@ void Preview::init()
resize(QSize(640, 480).expandedTo(minimumSizeHint()));
// buddies
searchLabel->setBuddy(searchTextLine);
searchLabel->setBuddy(searchTextCMB);
searchLabel->setText(tr("&Search for:"));
nextButton->setText(tr("&Next"));
@ -176,26 +268,25 @@ void Preview::init()
"RCL.SEARCH.PREVIEW");
// signals and slots connections
connect(searchTextLine, SIGNAL(textChanged(const QString&)),
this, SLOT(searchTextLine_textChanged(const QString&)));
connect(searchTextCMB, SIGNAL(activated(int)),
this, SLOT(searchTextFromIndex(int)));
connect(searchTextCMB, SIGNAL(editTextChanged(const QString&)),
this, SLOT(searchTextChanged(const QString&)));
connect(nextButton, SIGNAL(clicked()), this, SLOT(nextPressed()));
connect(prevButton, SIGNAL(clicked()), this, SLOT(prevPressed()));
connect(clearPB, SIGNAL(clicked()), searchTextLine, SLOT(clear()));
connect(clearPB, SIGNAL(clicked()), searchTextCMB, SLOT(clearEditText()));
connect(pvTab, SIGNAL(currentChanged(QWidget *)),
this, SLOT(currentChanged(QWidget *)));
connect(bt, SIGNAL(clicked()), this, SLOT(closeCurrentTab()));
m_dynSearchActive = false;
m_canBeep = true;
m_currentW = 0;
if (prefs.pvwidth > 100) {
resize(prefs.pvwidth, prefs.pvheight);
}
m_loading = false;
currentChanged(pvTab->currentWidget());
m_justCreated = true;
m_haveAnchors = false;
m_curAnchor = 1;
}
void Preview::closeEvent(QCloseEvent *e)
@ -273,11 +364,11 @@ bool Preview::eventFilter(QObject *target, QEvent *event)
} else if (m_dynSearchActive) {
if (keyEvent->key() == Qt::Key_F3) {
LOGDEB2(("Preview::eventFilter: got F3\n"));
doSearch(searchTextLine->text(), true, false);
doSearch(searchTextCMB->currentText(), true, false);
return true;
}
if (target != searchTextLine)
return QApplication::sendEvent(searchTextLine, event);
if (target != searchTextCMB)
return QApplication::sendEvent(searchTextCMB, event);
} else {
if (edit &&
(target == edit || target == edit->viewport())) {
@ -285,7 +376,7 @@ bool Preview::eventFilter(QObject *target, QEvent *event)
(keyEvent->key() == Qt::Key_F &&
(keyEvent->modifiers() & Qt::ControlModifier))) {
LOGDEB2(("Preview::eventFilter: got / or C-F\n"));
searchTextLine->setFocus();
searchTextCMB->setFocus();
m_dynSearchActive = true;
return true;
} else if (keyEvent->key() == Qt::Key_Space) {
@ -307,23 +398,27 @@ bool Preview::eventFilter(QObject *target, QEvent *event)
return false;
}
void Preview::searchTextLine_textChanged(const QString & text)
void Preview::searchTextChanged(const QString & text)
{
LOGDEB2(("search line text changed. text: '%s'\n", text.ascii()));
LOGDEB1(("Search line text changed. text: '%s'\n",
(const char *)text.toAscii()));
m_searchTextFromIndex = -1;
if (text.isEmpty()) {
m_dynSearchActive = false;
// nextButton->setEnabled(false);
// prevButton->setEnabled(false);
clearPB->setEnabled(false);
} else {
m_dynSearchActive = true;
// nextButton->setEnabled(true);
// prevButton->setEnabled(true);
clearPB->setEnabled(true);
doSearch(text, false, false);
}
}
void Preview::searchTextFromIndex(int idx)
{
LOGDEB1(("search line from index %d\n", idx));
m_searchTextFromIndex = idx;
}
PreviewTextEdit *Preview::currentEditor()
{
LOGDEB2(("Preview::currentEditor()\n"));
@ -351,8 +446,8 @@ void Preview::emitSaveDocToFile()
void Preview::doSearch(const QString &_text, bool next, bool reverse,
bool wordOnly)
{
LOGDEB(("Preview::doSearch: text [%s] txtlen %d next %d rev %d word %d\n",
(const char *)_text.toUtf8(), _text.length(), int(next),
LOGDEB(("Preview::doSearch: text [%s] idx %d next %d rev %d word %d\n",
(const char *)_text.toUtf8(), m_searchTextFromIndex, int(next),
int(reverse), int(wordOnly)));
QString text = _text;
@ -363,25 +458,19 @@ void Preview::doSearch(const QString &_text, bool next, bool reverse,
return;
}
if (text.isEmpty()) {
if (m_haveAnchors == false) {
if (text.isEmpty() || m_searchTextFromIndex != -1) {
if (!edit->m_plaintorich->haveAnchors()) {
LOGDEB(("NO ANCHORS\n"));
return;
}
// The combobox indices are equal to the search ugroup indices
// in hldata, that's how we built the list.
if (reverse) {
if (m_curAnchor == 1)
m_curAnchor = edit->m_plaintorich->lastanchor;
else
m_curAnchor--;
edit->m_plaintorich->prevAnchorNum(m_searchTextFromIndex);
} else {
if (m_curAnchor == edit->m_plaintorich->lastanchor)
m_curAnchor = 1;
else
m_curAnchor++;
edit->m_plaintorich->nextAnchorNum(m_searchTextFromIndex);
}
LOGDEB(("m_curAnchor: %d\n", m_curAnchor));
QString aname =
QString::fromUtf8(edit->m_plaintorich->termAnchorName(m_curAnchor).c_str());
QString aname = edit->m_plaintorich->curAnchorName();
LOGDEB(("Calling scrollToAnchor(%s)\n", (const char *)aname.toUtf8()));
edit->scrollToAnchor(aname);
// Position the cursor approximately at the anchor (top of
@ -440,14 +529,14 @@ void Preview::doSearch(const QString &_text, bool next, bool reverse,
void Preview::nextPressed()
{
LOGDEB2(("PreviewTextEdit::nextPressed\n"));
doSearch(searchTextLine->text(), true, false);
LOGDEB2(("Preview::nextPressed\n"));
doSearch(searchTextCMB->currentText(), true, false);
}
void Preview::prevPressed()
{
LOGDEB2(("PreviewTextEdit::prevPressed\n"));
doSearch(searchTextLine->text(), true, true);
LOGDEB2(("Preview::prevPressed\n"));
doSearch(searchTextCMB->currentText(), true, true);
}
// Called when user clicks on tab
@ -456,7 +545,6 @@ void Preview::currentChanged(QWidget * tw)
LOGDEB2(("PreviewTextEdit::currentChanged\n"));
PreviewTextEdit *edit =
tw->findChild<PreviewTextEdit*>("pvEdit");
m_currentW = tw;
LOGDEB1(("Preview::currentChanged(). Editor: %p\n", edit));
if (edit == 0) {
@ -470,7 +558,7 @@ void Preview::currentChanged(QWidget * tw)
connect(this, SIGNAL(printCurrentPreviewRequest()), edit, SLOT(print()));
edit->installEventFilter(this);
edit->viewport()->installEventFilter(this);
searchTextLine->installEventFilter(this);
searchTextCMB->installEventFilter(this);
emit(previewExposed(this, m_searchId, edit->m_docnum));
}
@ -507,7 +595,7 @@ PreviewTextEdit *Preview::addEditorTab()
void Preview::setCurTabProps(const Rcl::Doc &doc, int docnum)
{
LOGDEB1(("PreviewTextEdit::setCurTabProps\n"));
LOGDEB1(("Preview::setCurTabProps\n"));
QString title;
string ctitle;
if (doc.getmeta(Rcl::Doc::keytt, &ctitle) && !ctitle.empty()) {
@ -721,8 +809,6 @@ bool Preview::loadDocInCurrentTab(const Rcl::Doc &idoc, int docnum)
LoadGuard guard(&m_loading);
CancelCheck::instance().setCancel(false);
m_haveAnchors = false;
setCurTabProps(idoc, docnum);
QString msg = QString("Loading: %1 (size %2 bytes)")
@ -956,23 +1042,20 @@ bool Preview::loadDocInCurrentTab(const Rcl::Doc &idoc, int docnum)
// Position the editor so that the first search term is visible
m_haveAnchors = editor->m_plaintorich->lastanchor != 0;
if (searchTextLine->text().length() != 0) {
if (searchTextCMB->currentText().length() != 0) {
// If there is a current search string, perform the search
m_canBeep = true;
doSearch(searchTextLine->text(), true, false);
doSearch(searchTextCMB->currentText(), true, false);
} else {
// Position to the first query term
if (m_haveAnchors) {
QString aname =
QString::fromUtf8(editor->m_plaintorich->termAnchorName(1).c_str());
if (editor->m_plaintorich->haveAnchors()) {
QString aname = editor->m_plaintorich->curAnchorName();
LOGDEB2(("Call movetoanchor(%s)\n", (const char *)aname.toUtf8()));
editor->scrollToAnchor(aname);
// Position the cursor approximately at the anchor (top of
// viewport) so that searches start from here
QTextCursor cursor = editor->cursorForPosition(QPoint(0, 0));
editor->setTextCursor(cursor);
m_curAnchor = 1;
}
}
@ -989,14 +1072,15 @@ bool Preview::loadDocInCurrentTab(const Rcl::Doc &idoc, int docnum)
return true;
}
PreviewTextEdit::PreviewTextEdit(QWidget* parent,const char* name, Preview *pv)
: QTextEdit(parent), m_preview(pv), m_dspflds(false), m_docnum(-1)
PreviewTextEdit::PreviewTextEdit(QWidget* parent, const char* nm, Preview *pv)
: QTextEdit(parent), m_preview(pv),
m_plaintorich(new PlainToRichQtPreview()),
m_dspflds(false), m_docnum(-1)
{
setContextMenuPolicy(Qt::CustomContextMenu);
setObjectName(name);
setObjectName(nm);
connect(this, SIGNAL(customContextMenuRequested(const QPoint&)),
this, SLOT(createPopupMenu(const QPoint&)));
m_plaintorich = new PlainToRichQtPreview();
}
PreviewTextEdit::~PreviewTextEdit()

23
src/qtgui/preview_w.h
View file

@ -58,22 +58,27 @@ protected:
void mouseDoubleClickEvent(QMouseEvent *);
private:
PlainToRichQtPreview *m_plaintorich;
Preview *m_preview;
PlainToRichQtPreview *m_plaintorich;
bool m_dspflds;
string m_url; // filename for this tab
string m_ipath; // Internal doc path inside file
int m_docnum; // Index of doc in db search results.
// doc out of internfile (previous fields come from the index) with
// main text erased (for space).
Rcl::Doc m_fdoc;
// The input doc out of the index/query list
Rcl::Doc m_dbdoc;
// Saved rich (or plain actually) text: the textedit seems to
// sometimes (but not always) return its text stripped of tags, so
// this is needed (for printing for example)
QString m_richtxt;
Qt::TextFormat m_format;
// Temporary file name (possibly, if displaying image). The
// TempFile itself is kept inside main.cpp (because that's where
// signal cleanup happens), but we use its name to ask for release
@ -92,14 +97,14 @@ class Preview : public QWidget {
Preview(int sid, // Search Id
const HighlightData& hdata) // Search terms etc. for highlighting
: QWidget(0), m_searchId(sid), m_hData(hdata)
: QWidget(0), m_searchId(sid), m_searchTextFromIndex(-1), m_hData(hdata)
{
init();
}
~Preview(){}
virtual void closeEvent(QCloseEvent *e );
virtual bool eventFilter(QObject *target, QEvent *event );
/**
* Arrange for the document to be displayed either by exposing the tab
* if already loaded, or by creating a new tab and loading it.
@ -112,7 +117,8 @@ class Preview : public QWidget {
friend class PreviewTextEdit;
public slots:
virtual void searchTextLine_textChanged(const QString& text);
virtual void searchTextChanged(const QString& text);
virtual void searchTextFromIndex(int);
virtual void doSearch(const QString& str, bool next, bool reverse,
bool wo = false);
virtual void nextPressed();
@ -138,18 +144,17 @@ private:
int m_searchId;
bool m_dynSearchActive;
// Index value the search text comes from. -1 if text was edited
int m_searchTextFromIndex;
bool m_canBeep;
bool m_loading;
QWidget *m_currentW;
HighlightData m_hData;
bool m_justCreated; // First tab create is different
bool m_haveAnchors; // Search terms are marked in text
int m_lastAnchor; // Number of last anchor. Then rewind to 1
int m_curAnchor;
QTabWidget* pvTab;
QLabel* searchLabel;
QLineEdit* searchTextLine;
QComboBox *searchTextCMB;
QPushButton* nextButton;
QPushButton* prevButton;
QPushButton* clearPB;

16
src/qtgui/reslist.cpp
View file

@ -250,12 +250,22 @@ string QtGuiResListPager::iconUrl(RclConfig *config, Rcl::Doc& doc)
class PlainToRichQtReslist : public PlainToRich {
public:
virtual ~PlainToRichQtReslist() {}
virtual string startMatch() {
virtual string startMatch(unsigned int idx)
{
if (m_hdata) {
string s1, s2;
stringsToString<vector<string> >(m_hdata->groups[idx], s1);
stringsToString<vector<string> >(m_hdata->ugroups[m_hdata->grpsugidx[idx]], s2);
LOGDEB(("Reslist startmatch: group %s user group %s\n", s1.c_str(), s2.c_str()));
}
return string("<span class='rclmatch' style='color: ")
+ string((const char *)prefs.qtermcolor.toAscii()) + string("'>");
}
virtual string endMatch() {return string("</span>");}
virtual string endMatch()
{
return string("</span>");
}
};
static PlainToRichQtReslist g_hiliter;

1
src/qtgui/ssearch_w.h
View file

@ -38,7 +38,6 @@ public:
setupUi(this);
init();
}
~SSearch(){}
virtual void init();
virtual void setAnyTermMode();

88
src/query/plaintorich.cpp
View file

@ -49,13 +49,22 @@ static string vecStringToString(const vector<string>& t)
return sterms;
}
struct MatchEntry {
pair<int, int> offs;
unsigned int grpidx;
MatchEntry(int sta, int sto, unsigned int idx)
: offs(sta, sto), grpidx(idx)
{
}
};
// Text splitter used to take note of the position of query terms
// inside the result text. This is then used to insert highlight tags.
class TextSplitPTR : public TextSplit {
public:
// Out: begin and end byte positions of query terms/groups in text
vector<pair<int, int> > tboffs;
vector<MatchEntry> tboffs;
TextSplitPTR(const HighlightData& hdata)
: m_wcount(0), m_hdata(hdata)
@ -67,7 +76,7 @@ class TextSplitPTR : public TextSplit {
for (vector<vector<string> >::const_iterator vit = hdata.groups.begin();
vit != hdata.groups.end(); vit++) {
if (vit->size() == 1) {
m_terms.insert(vit->front());
m_terms[vit->front()] = vit - hdata.groups.begin();
} else if (vit->size() > 1) {
for (vector<string>::const_iterator it = vit->begin();
it != vit->end(); it++) {
@ -91,8 +100,9 @@ class TextSplitPTR : public TextSplit {
// pos, bts, bte));
// If this word is a search term, remember its byte-offset span.
if (m_terms.find(dumb) != m_terms.end()) {
tboffs.push_back(pair<int, int>(bts, bte));
map<string, unsigned int>::const_iterator it = m_terms.find(dumb);
if (it != m_terms.end()) {
tboffs.push_back(MatchEntry(bts, bte, (*it).second));
}
// If word is part of a search group, update its positions list
@ -114,13 +124,13 @@ class TextSplitPTR : public TextSplit {
virtual bool matchGroups();
private:
virtual bool matchGroup(const vector<string>& terms, int dist);
virtual bool matchGroup(unsigned int idx);
// Word count. Used to call checkCancel from time to time.
int m_wcount;
// In: user query terms
set<string> m_terms;
map<string, unsigned int> m_terms;
// m_gterms holds all the terms in m_groups, as a set for quick lookup
set<string> m_gterms;
@ -191,9 +201,12 @@ static bool do_proximity_test(int window, vector<vector<int>* >& plists,
return false;
}
// Find NEAR matches for the input group of terms, update highlight map
bool TextSplitPTR::matchGroup(const vector<string>& terms, int window)
// Find NEAR matches for one group of terms, update highlight map
bool TextSplitPTR::matchGroup(unsigned int grpidx)
{
const vector<string>& terms = m_hdata.groups[grpidx];
int window = m_hdata.groups[grpidx].size() + m_hdata.slacks[grpidx];
LOGDEB0(("TextSplitPTR::matchGroup:d %d: %s\n", window,
vecStringToString(terms).c_str()));
@ -203,26 +216,23 @@ bool TextSplitPTR::matchGroup(const vector<string>& terms, int window)
// A revert plist->term map. This is so that we can find who is who after
// sorting the plists by length.
map<vector<int>*, string> plistToTerm;
// For traces
vector<string> realgroup;
// Find the position list for each term in the group. Not all
// necessarily exist (esp for NEAR where terms have been
// stem-expanded: we don't know which matched)
// Find the position list for each term in the group. It is
// possible that this particular group was not actually matched by
// the search, so that some terms are not found.
for (vector<string>::const_iterator it = terms.begin();
it != terms.end(); it++) {
map<string, vector<int> >::iterator pl = m_plists.find(*it);
if (pl == m_plists.end()) {
LOGDEB0(("TextSplitPTR::matchGroup: [%s] not found in m_plists\n",
(*it).c_str()));
continue;
return false;
}
plists.push_back(&(pl->second));
plistToTerm[&(pl->second)] = *it;
realgroup.push_back(*it);
}
LOGDEB0(("TextSplitPTR::matchGroup:d %d:real group after expansion %s\n",
window, vecStringToString(realgroup).c_str()));
// I think this can't actually happen, was useful when we used to
// prune the groups, but doesn't hurt.
if (plists.size() < 2) {
LOGDEB0(("TextSplitPTR::matchGroup: no actual groups found\n"));
return false;
@ -261,15 +271,13 @@ bool TextSplitPTR::matchGroup(const vector<string>& terms, int window)
SETMINMAX(pos, sta, sto);
minpos = sto+1;
// Translate the position window into a byte offset window
int bs = 0;
map<int, pair<int, int> >::iterator i1 = m_gpostobytes.find(sta);
map<int, pair<int, int> >::iterator i2 = m_gpostobytes.find(sto);
if (i1 != m_gpostobytes.end() && i2 != m_gpostobytes.end()) {
LOGDEB0(("TextSplitPTR::matchGroup: pushing bpos %d %d\n",
i1->second.first, i2->second.second));
tboffs.push_back(pair<int, int>(i1->second.first,
i2->second.second));
bs = i1->second.first;
tboffs.push_back(MatchEntry(i1->second.first,
i2->second.second, grpidx));
} else {
LOGDEB(("matchGroup: no bpos found for %d or %d\n", sta, sto));
}
@ -284,22 +292,23 @@ bool TextSplitPTR::matchGroup(const vector<string>& terms, int window)
/** Sort integer pairs by increasing first value and decreasing width */
class PairIntCmpFirst {
public:
bool operator()(pair<int,int> a, pair<int, int>b) {
if (a.first != b.first)
return a.first < b.first;
return a.second > b.second;
bool operator()(const MatchEntry& a, const MatchEntry& b) {
if (a.offs.first != b.offs.first)
return a.offs.first < b.offs.first;
return a.offs.second > b.offs.second;
}
};
// Look for matches to PHRASE and NEAR term groups. Actually, we
// handle all groups as NEAR (ignore order).
// Look for matches to PHRASE and NEAR term groups and finalize the
// matched regions list (sort it by increasing start then decreasing
// length)
// Actually, we handle all groups as NEAR (ignore order).
bool TextSplitPTR::matchGroups()
{
for (unsigned int i = 0; i < m_hdata.groups.size(); i++) {
if (m_hdata.groups[i].size() <= 1)
continue;
matchGroup(m_hdata.groups[i],
m_hdata.groups[i].size() + m_hdata.slacks[i]);
matchGroup(i);
}
// Sort regions by increasing start and decreasing width.
@ -324,6 +333,7 @@ bool PlainToRich::plaintorich(const string& in,
{
Chrono chron;
m_hdata = &hdata;
// Compute the positions for the query terms. We use the text
// splitter to break the text into words, and compare the words to
// the search terms,
@ -346,8 +356,8 @@ bool PlainToRich::plaintorich(const string& in,
// Iterator for the list of input term positions. We use it to
// output highlight tags and to compute term positions in the
// output text
vector<pair<int, int> >::iterator tPosIt = splitter.tboffs.begin();
vector<pair<int, int> >::iterator tPosEnd = splitter.tboffs.end();
vector<MatchEntry>::iterator tPosIt = splitter.tboffs.begin();
vector<MatchEntry>::iterator tPosEnd = splitter.tboffs.end();
#if 0
for (vector<pair<int, int> >::const_iterator it = splitter.tboffs.begin();
@ -365,8 +375,6 @@ bool PlainToRich::plaintorich(const string& in,
int hadcr = 0;
int inindent = 1;
// Value for numbered anchors at each term match
int anchoridx = 1;
// HTML state
bool intag = false, inparamvalue = false;
// My tag state
@ -391,22 +399,20 @@ bool PlainToRich::plaintorich(const string& in,
// we are at or after a term match, mark.
if (tPosIt != tPosEnd) {
int ibyteidx = chariter.getBpos();
if (ibyteidx == tPosIt->first) {
if (ibyteidx == tPosIt->offs.first) {
if (!intag && ibyteidx >= (int)headend) {
*olit += startAnchor(anchoridx);
*olit += startMatch();
*olit += startMatch(tPosIt->grpidx);
}
anchoridx++;
inrcltag = 1;
} else if (ibyteidx == tPosIt->second) {
} else if (ibyteidx == tPosIt->offs.second) {
// Output end of match region tags
if (!intag && ibyteidx > (int)headend) {
*olit += endMatch();
*olit += endAnchor();
}
// Skip all highlight areas that would overlap this one
int crend = tPosIt->second;
while (tPosIt != splitter.tboffs.end() && tPosIt->first < crend)
int crend = tPosIt->offs.second;
while (tPosIt != splitter.tboffs.end() &&
tPosIt->offs.first < crend)
tPosIt++;
inrcltag = 0;
}

34
src/query/plaintorich.h
View file

@ -21,6 +21,7 @@
#include <list>
#include "hldata.h"
#include "cstr.h"
/**
* A class for highlighting search results. Overridable methods allow
@ -31,7 +32,7 @@
class PlainToRich {
public:
PlainToRich()
: m_inputhtml(false)
: m_inputhtml(false), m_eolbr(false), m_hdata(0)
{
}
@ -71,18 +72,35 @@ public:
);
/* Overridable output methods for headers, highlighting and marking tags */
virtual std::string header() {return snull;}
virtual std::string startMatch() {return snull;}
virtual std::string endMatch() {return snull;}
virtual std::string startAnchor(int) {return snull;}
virtual std::string endAnchor() {return snull;}
virtual std::string startChunk() {return snull;}
virtual std::string header()
{
return cstr_null;
}
/** Return match prefix (e.g.: <div class="match">).
@param groupidx the index into hdata.groups */
virtual std::string startMatch(unsigned int)
{
return cstr_null;
}
/** Return data for end of match area (e.g.: </div>). */
virtual std::string endMatch()
{
return cstr_null;
}
virtual std::string startChunk()
{
return cstr_null;
}
protected:
const std::string snull;
bool m_inputhtml;
// Use <br> to break plain text lines (else caller has used a <pre> tag)
bool m_eolbr;
const HighlightData *m_hdata;
};
#endif /* _PLAINTORICH_H_INCLUDED_ */

11
src/query/reslistpager.cpp
View file

@ -40,9 +40,14 @@ static const string cstr_hlfontcolor("<font color=\"blue\">");
static const string cstr_hlendfont("</font>");
class PlainToRichHtReslist : public PlainToRich {
public:
virtual ~PlainToRichHtReslist() {}
virtual string startMatch() {return cstr_hlfontcolor;}
virtual string endMatch() {return cstr_hlendfont;}
virtual string startMatch(unsigned int)
{
return cstr_hlfontcolor;
}
virtual string endMatch()
{
return cstr_hlendfont;
}
};
static PlainToRichHtReslist g_hiliter;